Composite Signals and Fourier Series

To approximate a square wave with frequency f and amplitude A, the

terms of the series are as follows:

Frequencies: f, 3f, 5f, 7f,

Signals with an integer multiple of a base frequency f are called
harmonics

Amplitudes of sine waves with these frequencies:

Frequency

Amplitude

4A/

3f

4A/3

5f

4A/5

7f

4A/7

Note that amplitudes of higher frequencies decrease rapidly.

1

Composite analog signals

Digital versus Analog Bandwidth

The analog bandwidth B is the difference between the
highest and the lowest frequency required to send a
composite signal.
Units: Hertz (Hz)
Recall that 1 Hz = 1 s-1
The digital bandwidth (bit rate) is based on the length of a
bit in terms of the time taken to send the bit, measured in
b/s.
The two types of bandwidth are proportional to each other.
The exact relationship depends on two factors:
The number of harmonics used in the analog signal
(affects the analog bandwidth)
The number of signal levels used to represent a bit.
3

Digital versus Analog Bandwidth

Example:
If the high and low amplitudes of a sine wave are used to
encode a 1 and a 0, two bits can be encoded per sine wave
period.

Suppose we want to send 6000 b/s.

One extreme: sending a long series of consecutive 0 (or 1) bits

requires a signal with unvarying amplitude (i.e. frequency is 0 Hz).

Other extreme: sending alternating 0 and 1 bits requires a signal

of 3000 Hz. => 1 Hz represents 2 b/s.

Analog bandwidth: B = 3000 Hz (3000 0)

Digital bandwidth (capacity): C = 6000 b/s

Relationship: B = C/2

Adding harmonics increases B, so B >= n/2

4

Nyquist formula
The previous discussion assumed 2 amplitude
levels used to encode bits.
General case: Nyquist bit rate formula
C = 2B log2 L
where L is the number of levels used to encode
bits.

Nyquists formula assumes noise-free

communication.

Shannons Formula
Claude Shannon examined the problem of
communication through a noisy channel.
The result is a limitation on the capacity of a
channel based on the amount of noise.
The formula:
S

C B log 2 1
N

Signal-to-noise ratio
The ratio S / N is the signal to noise ratio.
Units are often reported in bels, which is the
logarithm (base 10) of the ratio
Decibels: 10 times the above value
Example 35 dB = 3.5 bels

The ratio can also be expressed as a pure number:

S/N is 3162, meaning that the signal power is
3162 times the noise power.

Few things to remember

Using the calculator how to find
log2 (x) = ln (x) / ln (2)
log10 (x) = ln (x) / ln (10)
How to convert S/N to/from dB
y db = 10 log 10 (S/N)
10 y/10 => S/N
Also,
y = log n x

really means x = n y
8

Solving Questions using Nyquist Formula

C = 2B log2 L

1. Given C in bits/second and L (a number). Find B in Hz?

Example : C = 1 106 b/s and L = 8
1 106 = 2 B log2 (8)
B = 1.6 107 Hz

Solving Questions using Nyquist Formula

C = 2B log2 L

2. Given C in bits/second and B in Hz. Find L?

Example : C = 7 106 b/s and B = 1.75 106 Hz
7 106 = 2 (1.75 106) log2 (L)
L=4

10

Solving Questions using Nyquists Formula

C = 2B log2 L

3. Given B in Hz and L (in number). Find C in b/s?

Example : B = 6 106 Hz and L = 12
C = 2 (6 106) log2 (12)
C = 4.3 107 b/s

11

Solving Questions using Shannons Formula

S

C B log 2 1
N

1. Given C in b/s and B in Hz. Find S/N ?

Example : C = 5 106 b/s and B = 1.25 106
5 106 = (1.25 106 ) log2 (1+ S/N)
S/N = 16
Another variation is to find S/N in decibels.
S/N = 16 = log10 (16) = 1.20 bels
S/N = 1.20 * 10 = 12 decibels

12

Solving Questions using Shannons Formula

S

C B log 2 1
N

2. Given C in b/s and S/N. Find B in Hz?

Example : C = 1 109 Hz and S/N = 40 dB
First convert 40 dB to S/N ratio then apply Shannons
Formula
1 109 = B log2 (1 + 1.0104)
B = 75.25 106 Hz

13

Solving Questions using Shannons Formula

S

C B log 2 1
N

3. Given B in Hz and S/N. Find C in b/s?

Example : B = 4 109 Hz and S/N = 1023 (not in dB)
C = 4 109 log2 (1 + 1023)
C = 4 1010 b/s

14

Solving Questions using Shannons Formula

S

C B log 2 1
N

4. Given B in Hz and S/N in dB. Find C in b/s?

Example : B = 8 106 Hz and S/N = 45 dB
First convert decibels to S/N ratio => 10 40/10
C = 8 106 log2 (1 + 1.0 104)
C = 8 1010 1.33 101
C = 1.64 1012 b/s

15

Question 1
What is the maximum theoretical capacity in bits
per second of a coaxial cable band with a
frequency spectrum of 50 MHz to 100 MHz and a
signal to noise ratio of 40 dB?

16

Question 1
What is the maximum theoretical capacity in bits
per second of a coaxial cable band with a
frequency spectrum of 50 MHz to 100 MHz and a
signal to noise ratio of 40 dB?
Shannons formula:

C B log 2 1
N

First, determine S/N:

40 = 10 log10(S/N)
4 = log10(S/N)
1.0104 = S/N
Determine B: 1.0108 5.0107 = 5.0 107 b/s
17

Question 1
What is the maximum theoretical capacity in bits
per second of a coaxial cable band with a
frequency spectrum of 50 MHz to 100 MHz and a
signal to noise ratio of 40 dB?
S

C B log 2 1
N

Now, apply the formula:

C = 5.0107 log2(1+1.0104)
C = 5.0107 1.33101
C = 6.64108 b/s (or 664 Mb/s)

18

Question 2
We have a channel with a 1 MHz bandwidth. The
ratio S/N is 63. What is the appropriate bit rate
and number of signal levels?

19

Question 2
We have a channel with a 1 MHz bandwidth. The ratio S/N
is 63. What is the appropriate bit rate and number of signal
levels?
First, use Shannons formula to find the upper data rate at 1
signal level:

C B log 2 1
N

20

Question 2
We have a channel with a 1 MHz bandwidth. The ratio S/N
is 63. What is the appropriate bit rate and number of signal
levels?
First, use Shannons formula to find the upper data rate at 1
signal level:

C B log 2 1
N
C = 1106 log2(1+63)
= 1106 log2(64)
= 1106 6
= 6 106 b/s
21

Question 2
We have a channel with a 1 MHz bandwidth. The ratio S/N
is 63. What is the appropriate bit rate and number of signal
levels?
Next, use the Nyquist formula to obtain the number of
signal levels L:

C = 2B log2 L

22

Question 2
We have a channel with a 1 MHz bandwidth. The ratio S/N
is 63. What is the appropriate bit rate and number of signal
levels?
Next, use the Nyquist formula to obtain the number of
signal levels L:

C = 2B log2 L
6 106 = 2 1 106 log2(L)
3 = log2(L)
8=L
Therefore, 8 signal levels can be used to send bits at 6 10 6
b/s over this channel.
23

Question 3
Suppose that an FM radio station is allocated 200
KHz of bandwidth, and wants to digitally
broadcast stereo CD music which means that 16
bits need to be sent at a sample rate of 44,100
samples per second for each of 2 sound channels.
How many signal levels are needed?
C = 2B log2 L
44100x16x2 = 2Blog2L

24

Type of Questions
Given B and S/N ratio
- find C
- find L based on C
Given B, L
- find C
- find required S/N
- find S/N in dB

25